combined wind and waves over a fringing reef...2 modeling package, called twave (sánchez et al....
TRANSCRIPT
1
Combined Wind and Waves over a Fringing Reef
Alejandro Sánchez, Jane McKee Smith, Zeki Demirbilek, and Stan Boc
Coastal Hydraulics Laboratory, U.S. Army Corps of Engineers, Vicksburg MS
1. Introduction
Tropical islands may encounter strong wave and wind events due to tropical cyclones
which cause severe damage and life-threatening conditions. The capability to quickly
forecast the coastal wind and wave conditions for estimating the resulting coastal
inundation is crucial for planners and decision-makers responsible for early warnings,
evacuations and relief procedures (Cheung et al. 2003). Protective coastal structures are
commonly designed using hindcast wave and water levels and/or simulating a large
number of hypothetical cyclones representative of historical events (e.g. Thompson and
Scheffner, 2002).
One of the challenges in estimating storm-induced coastal inundation is the prediction of
the nearshore waves. This requires using nonlinear wave models such as Boussinesq-
type models to obtain accurate estimate of waves. These two-dimensional (2-D) fully
nonlinear models are computationally demanding. An alternative is to use simpler 1-D
models to simulate wave transformation over several transects along the coastlines that
are of concern for flood innundation. Nwogu (2006) was able to obtain similar wave
runup using 1-D and 2-D Boussinesq wave model simulations of nearshore waves during
Hurricane Iniki along the coast in Kauai, Hawaii. The 1-D models have been used
extensively and successfully in calculating waves in the surf zone (e.g., Gerritsen 1980;
Thornton and Guza 1983; Dally et al. 1985; Dally 1992; Larson and Kraus 1992; Massel
and Brinkman 2001), bottom morphology changes (e.g., Van Rijn et al. 2003) and
currents (e.g., Thornton and Guza 1986; Ruessink et al. 2001; Gourlay 2005).
Over the past decade, the Coastal and Hydraulics Laboratory (CHL) of the U.S. Army
Engineer Research and Development Center (ERDC), has developed and improved
numerical modeling procedures for determining coastal waves and inundation for
tropical storms (e.g., Thompson and Scheffner, 2002; Demirbilek et al. 2007a;
Demirbilek and Nwogu 2007b). The Surge and Wave Island Modeling Studies (SWIMS)
project of the US Army Corps of Engineers (USACE) is developing predictive capabilities
for island flood estimates. One of these tools is a simple, first-order, integrated wave
2
modeling package, called TWAVE (Sánchez et al. 2007). TWAVE is intended for the US
civil defense agencies and USACE District Offices, providing users a practice-oriented
package which is user-friendly, flexible and robust and suitable for the feasibility and
reconnaissance type engineering planning studies.
In this paper, the focus is on wave transformation over fringing reefs and potential
effects of wind on reef-top dynamics. In the first section, measurements from two
laboratory studies of wave transformation and setup over fringing reefs are discussed. In
the second section, existing empirical and semi-analytical formulations for estimating
wave setup over a fringing reef are reviewed. Revised equations are proposed that
include the contribution of wind setup. The numerical models results are compared with
laboratory measurements of wave heights, mean water levels and runup in the third
section of this paper. Conclusions and recommendations are provided at the end of
paper.
2.0 Laboratory Datasets
2.1 University of Michigan Dataset
The first experimental dataset use in this study is that of Demirbilek et al. (2007a).
These experiments consist of approximately 80 test carried out on a 1:64 scale model of a
fringing reef type profile in a wind-wave flume at the University of Michigan (UM). The
reef profile is similar to that of Seelig (1983) except it does not include a reef lagoon
(Figure 1). A similar profile was also used by and Thompson et al. (2005) that had a
more narrow reef width. Reefs with a flat reef top are common in the Southern coast of
the Island of Guam. The flume is 35 m long, 0.7 m wide and 1.6 m high. Random
unidirectional waves were generated using a plunger type wave maker and winds were
generated using a 40-horsepower blower capable of producing winds up to 30 m/sec.
The instrumentation consisted of nine capacitance-wire gauges to measure wave and
water surface elevations along the reef profile, one 1-m-long capacitance-wire runup
gauge, and two anemometers for the wind speeds. The significant wave height was
calculated using 4s oH m= where om is the integration of the water surface spectral
density over frequencies smaller than infragravity waves (<0.04 Hz). This data set is
referred to hereafter as the UM dataset.
3
0 200 400 600 800 1000 1200 1400 1600
-10
0
10
20
30
Distance, m
Dep
th, m
Figure 1. Experimental setup for UM fringing reef experiments. Dimensions are in
prototype scale. Depth is with reference to reef-top
The deepwater wave steepness, defined as /o oH L where oH and oL are the deep water
significant wave height and wave length respectively, in the UM dataset varied between
0.015-0.05. The laboratory results were scaled up to prototype scale of 1:64 using the
Froude scaling. The bottom friction and wind drag coefficients were not scaled. Since
the experiments were carried out over smooth bottoms, the bottom friction was
considered to be negligible. The wind drag coefficient was calibrated by adjusting its
value to fit laboratory measurements from tests which had no paddle generated waves
(wind only testa). In the numerical study presented here, STWAVE and ADCIRC models
were used to obtain a linear function of the drag coefficient as a function of the wind
speed. Nonetheless, even the wind-only test cases generated waves that contributed to
the wave setup over the reef. Therefore, the calibrated drag coefficients might be slightly
overestimated.
2.2 Hayman Island Dataset
The second laboratory dataset was obtained from the monochromatic wave experiments
conducted by Gourlay (1994) for an idealized fringing reef profile similar to those found
in the southern side of Hayman Island, North Queensland Australia (Figure 2). The
dataset consists of 18 tests of regular waves with 1-3.5 m heights, 3.8-6.8 sec periods, and
a deepwater wave steepness of approximately 0.048. The water depth over the reef
varied from 1.4 m to 3.4 m.
#1 #9 #4 #8
Wave Probes
1:5
1:18.8
1:10.6 1:12
Wedge Wavemaker
#7
4
0 100 200 300 400 500-8
-6
-4
-2
0
2
4
Distance (m)
Dep
th (m
)
Figure 2. Experimental setup for Hayman Island fringing reef experiments. Dimensions
are in prototype scale.
3.0 Empirical Relations
3.1 Waves Setup
Seelig (1983) obtained the following empirical relation for wave setup by using
laboratory experiments for a fringing reef profile similar to the one used in the UM
experiments. He related the wave setup to a parameter resembling the wave power as:
( )( )
210
210
0.92 0.77log for 0 m
1.25 0.73log for 2 m
o r
o r
H T h
H T hη
⎧− + =⎪∆ = ⎨− + =⎪⎩
(1)
where T is the wave period, oH is the deepwater wave height and rh is the still water
depth over the reef. One of the disadvantages of Equation 1 is that it is only valid for
depths 0 and 2 m. To obtain the wave setup at water depths other than 0 and 2 m, the
setup is usually interpolated. We should note that Equation (1) can produce negative
values of wave setup for small wave heights and periods.
Gourlay (1996a) investigated wave setup and wave-generated current over reefs using
physical modeling experiments for idealized fringing and platform reefs. He determined
that the setup increased with both increasing wave heights and period, and decreased
with increasing depth over the reef top. Using a wave energy balance equation, Gourlay
(1996b) obtained an implicit solution for wave setup.
Wave Probes Wedge Wavemaker
1:4.5
1:33.3 1:83.3 1:106.7
5
( )
1/ 2 22 2
3/ 23 11 4
64o r
p R rr
g H T hK K KT gh
ηη ππ η
⎡ ⎤+ ∆∆ = − −⎢ ⎥
+ ∆ ⎣ ⎦ (2)
where pK is a shape factor which accounts for the effects various dissipation
mechanisms over the entire reef profile, RK is a wave reflection factor, rK is a wave
transmission factor and g is the gravitational constant. This relation has been used
successfully for predicting wave setup on reefs with steep faces by Gourlay (1996b, 1997).
The K factors in Equation 2 must be calibrated for each reef profile. As pointed out by
Gourlay (1996b), the factors appearing in Equation 2 are not necessarily constant for
different reef profiles. The shape factor for example is expected to be a function of the
still-water depth and offshore wave height and period, since these will determine the
segment of reef (bathymetry) over which the waves will break. Because the bottom
friction is not included, Equation 2 is not expected to be accurate for wide, shallow reefs
with rough bottoms (Gourlay 1996b).
The first term in Equation 1 is related to the still-water depth (SWL) and the second to
the offshore wave energy. The setup increases as the SWL over the reef decreases
(Gerritsen 1980, Gourlay 1996ab, Gourlay 2005). The setup is also related to the
offshore wave steepness (Gerritsen, 1980), therefore we provide a expression similar to
that of Seelig (1983):
( )2 2100.18 0.48log 5.53 /wave r o oh H T H Tη∆ = − + − (3)
Equation 3 along with Seelig’s and Gourlay’s equations are compared to the measured
setup over the reef top in Figure 3. One of the problems with Seelig’s empirical
formulation is that it can lead to negative wave setup values for small wave height
conditions. Gourlay’s equation was calibrated by assuming that the wave reflection is negligible ( RK =0) and adjusting pK to match the measurements. Demirbilek et al.
(2007a) calculated reflection coefficients for the wind-wave flume experiments of less
than 10%. According to Gourlay (1996b), the wave reflection does not have a significant
effect on wave setup when the reflection coefficient is less than 20%.
6
0 0.5 1 1.5 2
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
Measured wave setup, m
Cal
cula
ted
wav
e se
tup,
m
PresentSeelig 1983Gourlay 1994, Kp=0.13, KR=0
Figure 3. Comparison of measured (UM dataset) and computed wave-setup using Seelig
(1983), Gourlay (1994) and a modified form of Seelig’s equation.
Equation 3 shows improved results compared to both Seelig’s and Gourlay’s
formulations for the UM dataset. Equation 3 also does not produce negative setup
values. Although Equation 3 does not have any free calibration parameters, the
empirical constants are expected to be a function of the reef geometry, and therefore
Equation 3 is only valid for fringing reef type profiles.
3.2 Wind Setup
The above results show that it is important to consider the wind setup in empirical
estimates of water levels over a fringing reef. The following empirical relation, which is
based on the momentum balance of wind stress and water surface gradient provides a
simple expression to estimate the wind contribution to the setup over a fringing reef:
2
( )d
windr wave
WC UK h
ηη
∆ =+ ∆
(4)
where W is the width of the reef top in meters, dC is the wind drag coefficient, U is the
wind speed in m/sec, K is an empirical coefficient, and rh is the still-water depth over
the reef top in meters and waveη∆ is the wave setup calculated using Equation 3. The
total water level over the reef top is then total wave windη η η∆ = ∆ + ∆ . The coefficient K in
the denominator of Equation 4 was found by minimizing the root-mean-squared error
7
(RMSE), however the results are not sensitive to this parameter and reasonable results
were found for values of K between 800,000 and 120,000.
0 0.5 1 1.50
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
Measured setup, m
Cal
cula
ted
setu
p, m
Figure 4. Comparison of measured (UM dataset) and calculated setup over the reef top.
● Only wave setup (Equation 3); ▪ Wind and Wave setup (Equations 3 and 4)
It is noted that both empirical Equations (3) and (4) were obtained from the UM dataset.
These empirical formulas are applicable only to fringing reef profiles with steep faces.
Additional validation data are required to determine the applicability of these proposed
formulas to other reef geometries different than that of Guam used in the UM
experiments.
4.0 Numerical Models
4.1 STWAVE and ADCIRC
The STeady-state spectral WAVE (STWAVE) model (Smith et al. 2001) was used to
transform waves over the reef. The ADvanced CIRCulation (ADCIRC) model (Luettich et
al. 1992) was used for water level estimates over the reef. STWAVE was run using a 5-m
grid resolution. ADCIRC was run using a 10- to 30-m variable resolution grid and a
time-step of 0.2 sec until steady-state conditions were achieved. A two-way coupling
between the models was performed by passing the wave radiation stress gradients from
STWAVE to ADCIRC and then re-running STWAVE with adjusted depths to account for
8
the mean water surface elevations from ADCIRC. This process was repeated at least one
time and then STWAVE was re-run a final time. For these tests, the wave breaking was
modified in STWAVE to apply the Battjes and Jannsen (1978) formulation given by
214b b bD gBfH Qρ= (5)
where bD is the energy dissipation per unit area, ρ is water density, B is the wave
breaking intensity factor (set to 1.0), f is the wave frequency, Hrms is the root-mean-
square (rms) wave height, d is the total water depth and bQ is the fraction of broken
waves given by the implicit relation 2 2ln (1 )rms b b bH Q H Q= − . For steep slopes this
formulation does not produce enough wave dissipation to counter wave shoaling
(Jannsen and Battjes 2007). To remediate this, the significant wave height was limited to
the breaking wave height given by Miche’s criterion (1951) 0.88 tanh( ) /bH kd k= , where
k is the wave number.
The finite element circulation model ADCIRC was forced with the wave radiation stress
gradients from STWAVE and wind stress wx d aC U Uτ ρ= where aρ is the air density,
dC is the wind drag coefficient and U is the average wind speed measured in the flume.
The wind drag coefficient was calibrated by matching the computed wind setup with
measured setup for the tests which had no waves or very small waves. For brevity, the
coupled STWAVE and ADCIRC modeling system will be hereafter referred to as STAD.
4.2 One-Dimensional Wave Energy Balance Model
With the assumption of alongshore uniformity in the bathymetry, the time-averaged
wave energy balance equation becomes
( cos ) 0g b fEC D Dx
θ∂+ + =
∂ (6)
where E is the wave energy per unit area, gC is the wave group velocity, θ is the wave
incidence angle, bD and fD are the wave energy dissipation due to breaking and bottom
friction respectively. Since the laboratory tests were conducted over a smooth surface,
the friction is considered to be negligible and set to zero.
Two types of formulations were used to estimate the breaking wave dissipation. The first
is the empirical model of Dally et al. (1985) is expressed as
min( , )b g s gD EC E E Cdκ ⎡ ⎤= −⎣ ⎦ (7)
9
where κ is an empirical decay coefficient and sE is the stable wave height energy equal
to
21 ( )8sE g dρ= Γ (8)
where Γ defines the stable wave height are percentage of the water depth. Following
Smith et al. (1993), we adapted 0.15 and 0.42 for the values of κ and Γ . This approach
has successfully used in modeling wave transformation over irregular bathymetry
including reefs (e.g., Gerritsen 1980, Dally 1992). Assuming that the waves follow a
Rayleigh distribution
22( ) exp
rms rms
H Hp HH H
⎡ ⎤⎞⎛= ⎢− ⎥⎟⎜
⎢ ⎥⎝ ⎠⎣ ⎦, (9)
the root-mean-squared (rms) wave height can be obtained from
2 2
0
1 1( )8 8 rmsE g H p H dH gHρ ρ
∞
= =∫ (10)
and is related to the significant wave height by 2s rmsH H= .
It is important to mention that Equation 7 does not depend on the chosen wave height
distribution. The wave height distribution is only used to obtain a characteristic wave
height from the wave energy. This means that Equation 7 can be used for both random
and monochromatic waves.
The second formulation adopted for wave breaking dissipation was based on the early
work of Battjes and Janssen (1978). Recently, Alsina and Baldock (2007) and Janssen
and Battjes (2007) separately obtained the following expression for wave breaking
dissipation for an unsaturated surf zone as
33
16rms
b bHD gBf Q
dπ ρ= (11)
where the fraction of broken waves is equal to
( )3 24 31 exp erf ( )23bQ R R R R
π⎞⎛= + + − −⎜ ⎟
⎝ ⎠ (12)
In Equation 12, erf is the error function and /b rmsR H H= . The breaking wave height is
obtained from 0.88 tanh( / 0.88) /bH kd kγ= , where, the breaker index is calculated as in
Alsina and Baldock (2007) using the expression of Battjes and Stive (1985) / 0.5 0.4 tanh(33 / )b o oH d H Lγ = = + . Equation 11 is obtained by following a similar
10
approach as in Battjes and Janssen (1978) except that rather then assuming a clipped
Rayleigh distribution, a complete Rayleigh distribution has been assumed.
4.3 Momentum balance over a reef
Assuming uniformity in the alongshore direction (y-direction) and ignoring mixing and
cross-shore currents, the time-averaged and depth-integrated 1D momentum equation
may be written as
( ) xxwx
Sg hx xηρ η τ∂ ∂
+ + =∂ ∂
(13)
where η is the wave setup relative to still-water level, h is the still-water depth relative
to a specific vertical datum, and xxS is the wave radiation stress in the x-direction and
wxτ is the wind stress in the x-direction. The wave radiation stress is approximated using
linear wave theory for an arbitrary wave angle as (Longuet-Higgins and Stewart 1964) as 2( (cos 1) 0.5)xxS E n θ= + − where θ is the incident wave angle and
[ ]0.5 1 2 / sinh(2 )n kd kd= + . Equations 6 and 13 are solved simultaneously from deep to
shallow water. When the formulation for wave breaking of Dally et al. (1985) is used in
the calculations, the model is referred hereafter as DDD85, and when the formulations of
Alsina and Baldock (2007) and Janssen and Battjes (2007) are used, the model is
designated by ASJB07.
5.0 Comparison of Models with the UM Data
The STAD, ABJB07 and DDD85 models are compared with the UM laboratory
measurements. The effects of wind on wave transformation and setup are shown for
subsets of the laboratory measurements.
5.1 Waves Only Tests
Figure 5 shows an example of measurements and computed significant wave heights and
mean water surface elevation for Tests 17 and 47 which were forced only with waves
(Demirbilek et al. 2007a). Gauge 4 (fourth measurement station from left end of the
flume) shows an obvious bias due to problems with the gauge (Demirbilek et al. 2007a)
but is included here for completeness. Wave setup is slightly underestimated by STAD
over the reef top. Significant wave height values are larger than measurements and those
obtained from ABJB07. The ABJB07 and DDD85 models predicted similar mean water
levels that are in good agreement with measured values. The STAD and ABJB07 models
have similar wave breaking parameterizations that produced nearly the same wave
11
heights except over the reef top where STAD produced slightly higher wave heights. The
largest difference between DDD85 and the other models is in the energy dissipation ,
which tends to zero once the wave height has reached its stable value. Consequently, this
would cause an overprediction of wave heights at the Gauges 8 and 9 for two test
conditions shown in Figure 5. DDD85 also predicts comparatively higher breaking wave
heights than the other two models.
-0.5
0
0.5
1
η, m
0
2
4
6
Hs, m
Test-17
800 1000 1200 1400 1600
0102030
Distance, m
Elev
atio
n, m
-0.5
0
0.5
1η
, m0
1
2
3
4
Hs, m
Test-47
800 1000 1200 1400 1600
0102030
Distance, m
Elev
atio
n, m
Figure 5. Comparison between measured and computed significant wave heights and mean water levels for Test 17 ( oH =4.99 m, pT =12.0 sec, rh =3.26 m, U =0.5 m/s) and
Test 47 ( oH =3.20 m, pT =12.0 sec, rh =1.92 m, U =0.7 m/s).
— ABJB07, — DDD85, — STAD.
Figure 6a shows the calculated versus measured significant wave heights for all the tests
which were forced with only waves using STAD. This approach predicts well the wave
heights up to Gauge 6 (just before reef rim) and underestimates wave heights at Gauge 7.
A possible cause of the underestimation is the wave breaking formulation. Gauge 4 data
points to instrumentation problems (Demirbilek et al. 2007a). The measured and STAD
calculated water surface elevations are compared in Figure 6b. There is a significant
12
underestimation of the both the wave setdown (Gauges 5-6) and setup (Gauges 7-9). In
addition, the wave setup is limited to 0.4 m over the reef.
0 2 4 60
1
2
3
4
5
6
7STAD
Measured Hs, m
Cal
cula
ted
H s, m
+50%
-50%
-0.5 0 0.5 1 1.5-0.5
0
0.5
1
1.5STAD
Measured η, m
Cal
cula
ted
η, m
+50%
-50%
Gauge No.
123456789
Figure 6. Comparison between measured and calculated significant wave heights and
average water levels for the events with only wind using STAD.
The measured and ABJB07 model calculated significant wave heights are shown in
Figure 7a. The agreement is good at all gauges (discarding gauge 4 data for reasons
previously stated). The measured and ABJB07 model calculated surface elevations are
compared in Figure 7b. Predicted wave setup over the reef top is within 50% of the
measured values, and wave setdown at Gauges 5 and 6 are underestimated.
0 2 4 60
1
2
3
4
5
6
7ABJB07
Measured Hs, m
Cal
cula
ted
Hs, m
+50%
-50%
-0.5 0 0.5 1 1.5-0.5
0
0.5
1
1.5ABJB07
Measured η, m
Cal
cula
ted
η, m
Gauge No.
+50%
-50%
123456789
b.
b. a.
a.
13
Figure 7. Comparison between measured and calculated significant wave heights (a) and
average water levels (b) for the events with only wind using ABJB07.
The DDD85 model predicted wave heights over the reef face are also within 50% of the
measurements (Figure 8a). Wave heights at the reef edge (Gauge 7) are underestimated
and overpredicted at Gauges 8 and 9. Difference between the predicted and data over
the reef top for some test conditions is greater than 50% of the measured values. Both
ABJB07 and DDD85 models predict the wave setup over the reef top but fail to match
wave setdown at Gauges 5 and 6..
0 2 4 60
1
2
3
4
5
6
7DDD85
Measured Hs, m
Cal
cula
ted
Hs, m
+50%
-50%
-0.5 0 0.5 1 1.5-0.5
0
0.5
1
1.5DDD85
Measured η, m
Cal
cula
ted
η, mGauge No.
+50%
-50%
123456789
Figure 8. Comparison between measured and calculated wave heights (a) and average
water levels (b) for the events with only wind using DDD85.
5.2 Tests with Waves and Wind
Figure 9 shows the comparison of measured and computed significant wave heights and
mean water levels for Tests 75 and 85 with both wind and waves present. Test 75 shows
a significant increase in wave height from Gauge 8 to 9 caused by the strong winds of
37.5 m/sec. The DDD85 model over predicts the wave heights on the reef face.
When wave reach the stable wave height limit in the case of DDD85 model, the wave
height remains constant (without the presence of friction). Consequently, the wave
setup does not change. In the ABJB07 model, a fraction of waves are breaking and this
causes an increase in the wave setup over the reef top.
.
b. a.
14
-1
0
1
2
η, m
0
1
2
3H
s, mTest-75
800 1000 1200 1400 1600
0102030
Distance, m
Elev
atio
n, m
-2
0
2
4
η, m
0
2
4
6
8
Hs, m
Test-85
800 1000 1200 1400 1600
0102030
Distance, m
Elev
atio
n, m
Figure 9. Comparison between measured and computed significant wave heights and mean water levels for Test 75 ( oH =2.18 m, pT =8.0 sec, rh =1.92 m, U =32.4 m/sec) and
Test 85 ( oH =5.06 m, pT =16.0 sec, rh =0.0 m, U =41.4 m/sec).
— ABJB07, — DDD85, — STAD.
Gauges 2 and 3 occurs for all test conditions. Wave heights over the reef top are
underestimated by STAD model especially for tests performed with shallow depths over
the reef-top; see the points at the bottom portion of Figure 7a. There is not a significant
difference in the wave height comparison with observations as compared to the waves
only tests. The wind contribution in the STAD helps to obtain better prediction of the
water surface elevations over the reef top (Figure 10b). There is a large scatter as
compared to wave only tests, and the bias in water level setup is less. In general, large
scatter occurs in wind-induced setup for tests with both wind and waves. The bias for
combined wind and waves setup is less.
15
0 2 4 60
1
2
3
4
5
6
7STAD
Measured Hs, m
Cal
cula
ted
H s, m+50%
-50%
-1 0 1 2 3 4-1
0
1
2
3
4STAD
Measured η, m
Cal
cula
ted
η, m
+50%
-50%
Gauge No.
123456789
Figure 10. Comparison between measured and calculated significant wave heights for the
events with both wind and waves using STAD.
Similar to the tests with only wave forcing, the ABJB07 model accurately predicted wave
heights over the reef face (Gauges 4-6) and at Gauges 7 and 8 for combined wind and
wave tests (Figure 11a). This could be partly due the over prediction of the setup over the
reef top (Figure 11b). The over prediction of setup on reef top is caused by higher deep
water waves, and if the offshore wave heights were reduced, the setup and wave heights
at Gauges 7, 8 and 9 would all decrease. A poor agreement between predicted and
measured wave heights is apparent at Gauge 9. Decreasing offshore wave heights would
further reduce this agreement in the wave heights. The addition of wind increased the
scatter in the water surface elevations as compared to the wave only tests.
0 2 4 60
1
2
3
4
5
6
7ABJB07
Measured Hs, m
Cal
cula
ted
Hs, m
+50%
-50%
-1 0 1 2 3 4-1
0
1
2
3
4ABJB07
Measured η, m
Cal
cula
ted
η, m
Gauge No.
+50%
-50%
123456789
b. a.
16
Figure 11. Comparison between measured and calculated significant wave heights (a) and
average water levels (b) for the events with both wind and waves using ABJB07.
Both the STAD and DDD85 models show clustering of wave heights as a function of
water depth over the reef. Since the clustering occurs also for wave only tests, this would
suggest that models are limiting the calculated wave heights according to the water
depth. There is an increase in the scatter between DDD85 model predicted and
measured water surface elevations (Figure 12b). Both ABJB07 and DDD85 model
overestimate water levels at Gauges 7 and 8, and underestimate water levels at Gauge 9.
0 2 4 60
1
2
3
4
5
6
7DDD85
Measured Hs, m
Cal
cula
ted
Hs, m
+50%
-50%
-1 0 1 2 3 4-1
0
1
2
3
4DDD85
Measured η, m
Cal
cula
ted
η, m
Gauge No.
+50%
-50%
123456789
Figure 12. Comparison between measured and calculated wave heights (a) and average
water levels (b) for the events with both wind and waves using DDD85.
5.3 Wind Only Tests
Figure 13 shows an example of a test case forced only by wind. This example illustrates
the importance of the wind in the momentum balance in the nearshore region. Even
though there was no paddle-driven wave forcing, relatively wave heights were recorded
at all gauges due to wind generation of waves in the flume. As expected all three models
produce almost identical average water surface elevations (assuming wind forcing only).
b. a.
17
-0.1
0
0.1
0.2
η, m
0
0.2
0.4
0.6
0.8H
s, mWind-cal4-U3
800 1000 1200 1400 1600
0102030
Distance, m
Elev
atio
n, m
-0.5
0
0.5
1
1.5
η, m
0
0.5
1
1.5
2
2.5
Hs, m
Wind-cal4-U8
800 1000 1200 1400 1600
0102030
Distance, m
Elev
atio
n, m
Figure 13. Comparison between measured and computed wave heights and mean water levels for test Wind-cal4-U3 ( oH =0.0 m, pT =0.0 sec, rh =1.02 m, U =18.5 m/sec) and
Wind-cal4-U8 ( oH =0.0 m, pT =0.0 sec, rh =1.02 m, U =41.2 m/sec).
— ABJB07, — DDD85, — STAD.
The measured and calculated (only ADCIRC run) water surface elevations for wind only
tests are compared in Figure 14a. The ADCIRC predicted values agree better with data in
deepwater and for strong wind speeds, but model instabilities developed near the reef
rim and continued in shallow depths on reef top for weak wind speeds. Since with the
presence of waves The identical results from ABJB07 and DDD85 models are combined
in Figure 18 that show water elevations are virtually the same as the ADCIRC model
results except for one test in which the 1D model predicted dry conditions at Gauges 8
and 9.
18
-0.5 0 0.5 1 1.5 2-0.5
0
0.5
1
1.5
2ABJB07 and DDD85
Measured η, m
Cal
cula
ted
η, m
Gauge No.
+50%
-50%
123456789
-0.5 0 0.5 1 1.5 2-0.5
0
0.5
1
1.5
2STAD
Measured η, m
Cal
cula
ted
η, m
+50%
-50%
Figure 14. Comparison between measured and calculated average water levels for the
events with only wind using STAD (a) and using ABJB07 and DDD85.
5.4 Summary
The performances of three different numerical model are evaluated relative to the deep-
water significant wave height Ho for predictive purposes. The relative percent error is
defined as meas comp ox x H− where the subscripts meas and comp stand for measured
and computed and x is either the significant wave height or the mean water level. The maximum relative percent errors for both the significant wave heights at Gauges 4 through 9 are shown in Figure 15.
b. a.
19
Hs Max. Relative Errors for Wave Only Tests
0
10
20
30
40
50
60
4 5 6 7 8 9
Gauge
%
STADABJB07DDD85
Hs Max. Relative Errors for Wind and Wave Tests
0
10
20
30
40
50
60
4 5 6 7 8 9
Gauge
%
STADABJB07DDD85
Figure 15. Significant wave height relative errors for the wave only tests (a), and wind and waves tests (b).
η Max. Relative Errors for Waves Only Tests
05
101520253035
4 5 6 7 8 9
Gauge
%
STADABJB07DDD85
η Max. Relative Errors for Waves and Wind Tests
05
101520253035
4 5 6 7 8 9
Gauge
%
STADABJB07DDD85
Figure 16. Significant wave height relative errors for the wave only tests (a), and wind and waves tests (b).
b. a.
b. a.
20
meas comp
1
1i iN
i o
x xN H
ε=
−= ∑ (13)
where the subscripts meas and comp stand for measured and computed, x is either the significant wave height or the mean water level, i indicates the i-th experiment of N number of experiments.
compared in Table 1 using the Root-Mean-Squared-Error (RMSE). The ABJB07 model
produced the best results for wave setup for both waves only and wind and waves
combined. The coupled STWAVE and ADCIRC models produced the least accurate
results for both waves only and wind and waves combined. This is likely due to the fact
that the wave and setup are not solved simultaneously and only two iterations were used.
Also, the resolution was somewhat courser. The accuracy of predictions by all three
model decreased for tests having both wind and waves.
Table 1. Evaluation of model performance against measured significant wave height
(Hs) and average water level (η), by comparison of the Root-Mean-Squared-Error
(RMSE) for three models
Dataset All Only Waves Waves and Wind Only Wind
Model Hs, m η, m Hs, m η, m Hs, m η, m η, m
STAD 0.620 0.268 0.548 0.120 0.673 0.344 0.244
ABJB07 0.559 0.237 0.542 0.063 0.572 0.314 0.179
DDD85 0.650 0.238 0.631 0.064 0.665 0.316 0.179
6.0 Runup
The empirical relation of Mase (1989) for the 2% runup (runup exceeded 2% of the time)
for gentle slopes is given by 0.71
2% 1.86R ξ= (15)
where ξ is the surf similarity parameter defined as tan / /o oH Lξ θ= . On fringing
reefs, waves usually break near the reef rim, may reform again over the reef top. Thus, it is logical to use the wave height over the reef top rH in Equation (15) instead of the deep
21
water wave height. The wave length in the surf similarity parameter is based on peak
period of waves. The water surface elevations on reef top are dominated by long period
infra-gravity waves, which cannot be represented with models discussed in this paper.
Therefore, the deepwater wave length had to be used in Equation (15). Figure 15 shows
comparison between measured runup versus estimates from Equation (15) obtained with
models ABJB07 and DDD85. The computed and measured runup values are in good agreement; calculated runup values are higher for smaller ( 2%R <1.5 m) runup values.
The model ABJB07 underestimates higher runup values since wave dissipation over the
reef top is not included in the DDD85.
0 1 2 3 4 50
1
2
3
4
5ABJB07
Measured R2%, m
Cal
cula
ted
R 2%, m
0 1 2 3 4 50
1
2
3
4
5DDD07
Measured R2%, m
Cal
cula
ted
R 2%, m
Figure 15. Comparison between measured and calculated 2% runup using Equation 15
with wave heights from ABJB07 (a) and DDD85 (b). Runup heights are respect to still-
water level.
7.0 Comparison with Hayman Island Dataset
In this section, estimates from wave energy balance models ABJB07 and DDD07 are
compared to the laboratory experiments of Gourlay (1994) (see Section 2.2). Measured
and computed wave heights and mean water levels across the fringing reef are depicted
in Figures 16 and 17. The model estimates of wave heights over the reef are fairly good;
there is considerable scatter in the measurements. Both models ABJB07 and DDD85
overestimate wave setup for Tests 1.1 and 2.1 (Figure 16). Agreement is fair for Test 2.4
a. b.
22
(Figure 17a). Seelig (1983) found that random waves of a given offshore significant wave
height and period produce less setup as compared to monochromatic waves of the same
wave height and period. Seelig attributed this difference in setup to the fact that
irregular waves contain about half the energy of equivalent monochromatic waves.
Based on this observation, the model ABJB07 should underestimate wave setup,.
100 200 300 4001
2
3
4
H, m
Test 1.1
100 200 300 400-0.2
0
0.2
0.4
η, m
100 200 300 400
-202468
Distance, m
Wat
er D
epth
, m
100 200 300 4000
2
4
H, m
Test 2.1
100 200 300 400-0.2
00.20.4
η, m
100 200 300 400
-20246
Distance, m
Wat
er D
epth
, m
Figure 16. Comparison between measured and computed wave heights and average water levels for Test 1.1 (a) ( iH =3.44 m, T =6.8 sec, rh =3.4 m) and 2.1 (b) ( iH =4.22 m,
T =10.0 sec, rh =1.4 m). — ABJB07, — DDD85
23
100 200 300 4000
1
2
3H
, m
Test 2.4
100 200 300 400-0.2
0
0.2
0.4
η, m
100 200 300 400
-20246
Distance, m
Wat
er D
epth
, m
100 200 300 4000
0.5
1
1.5
H, m
Test 3.6
100 200 300 400-0.05
0
0.05
η, m
100 200 300 400
-20246
Distance, mW
ater
Dep
th, m
Figure 17. Comparison between measured and computed wave heights and average water levels for Test 2.4 (a) ( iH =1.90 m, T =5.4 sec, rh =1.4 m) and 3.6 (b) ( iH =0.99 m,
T =3.8 sec, rh =2.1 m). — ABJB07, — DDD85
Measured and computed wave heights and mean water levels obtained with models
ABJB07 and DDD85 are compared in Figure 19. The results of DDD85 agree better with
the data for wave heights and mean water levels. The agreement is evident by the RMSE
values, rms errors are 0.6887 and 0.5645 respectively for wave heights and mean water
levels for model ABJB07, and for DDD85 rms errors are 0.3411 m and 0.3360 m
respectively.
24
0 1 2 3 40
0.5
1
1.5
2
2.5
3
3.5
4
Measured H, m
Cal
cula
ted
H, m
ABJB07
0 0.2 0.4-0.1
0
0.1
0.2
0.3
0.4
0.5
Measured η, m
Cal
cula
ted
η, m
ABJB07
Figure 18. Comparison between measured and computed wave heights (a) and mean
water levels (b) for ABJB07
0 1 2 3 40
0.5
1
1.5
2
2.5
3
3.5
4
Measured H, m
Cal
cula
ted
H, m
DDD85
0 0.2 0.4-0.1
0
0.1
0.2
0.3
0.4
0.5
Measured η, m
Cal
cula
ted
η, m
DDD85
Figure 19. Comparison between measured and computed wave heights (a) and mean
water levels (b) for DDD85
25
8.0 Discussion
The most obvious effect of the wind in the UM experiments is an apparent increase in the
water level on reef top. For an onshore wind speed of the order of 10 m/s, Van Rijn and
Wijnberg (1996) obtained a significant wind setup of about 0.25 m at the shoreline of a
steep barred beach in Lake Michigan. Dally (1992) observed that offshore winds have a
significant influence on surf zone dynamics by delaying the onset of wave breaking.
Onshore winds are commonly known by suffers to induce early wave breaking. This
impact is not included in the models discussed here. The laboratory measurements show
that the wind setup across the reef platform give rise to larger wave heights on the reef
top.
Consistent with Equation 13, the wave setup was found to increase as a function of the
difference between the offshore wave height and water depth over the reef. When the
waves began to break further offshore, less setup is produced. For example, by
increasing the breaking intensity factor B, more energy is lost in deep water causing a
smaller wave setup over the reef. Also, consistent with Equation 13, the wave setup
increases with decreasing water depth over the reef and increases with wind speed.
Breaking processes over fringing reefs differ from sandy beaches. The breaking intensity
is expected to be greater for fringing reefs because of their steep faces, high roughness
and shallow depths. Wave attenuation rates over reefs are commonly in the range of 75-
95% (e.g., Lugo-Fernandez et al. 1995; Brander et al. 2004). For the Alsina and Baldock
(2007) formulation for wave breaking dissipation, the parameter B (set to 1.0 here)
controls the breaking intensity. In the case of the wave breaking formulation of Dally et
al. (1985) the equivalent parameter is κ (set to 0.17 here). These parameters are
expected to be larger for fringing reefs. For example, further of the testing of the Alsina
and Baldock (2007) formulation revealed that B = 1.26 produces the best agreement with
the UM dataset. The parameters κ and B are expected to be a function of the breaker
type. The energy dissipation due to wave breaking was also found to be inversely
proportional to the incident wave period. The energy decay coefficient could be
parameterized as a function of the wave period to include this effect. The wave energy
decay coefficient is also expected to be a function of the breaker type (reefs are more
likely to induce plunging breakers and beaches, spilling breakers).
The other free parameters in these wave breaking formulations which can be calibrated
to some extent are the breaking wave height Hb in Alsina and Baldock (2007) and the
26
stable wave height parameter Γ in Dally et al. (1985). These parameters control the
initiation and cessation of wave breaking and therefore the saturated wave heights.
Laboratory measurements suggest that a stable wave height parameter Γ based on a
saturated surf zone for sand beaches (Γ=0.42 ) and the breaking wave height according
the Miche’s criterion are too small for fringing reefs.
More research is needed on how to select the free parameters in the wave breaking
formulations to account for different breaker types, bottom conditions and possibly wind
conditions. Nonetheless, the current results show that reasonable estimates of wave
height and water levels can be obtained over fringing reefs without any calibration.
The reason the STWAVE and ADCIRC models under predicted setup over the fringing
reef is related to the coupling between the models and the wave breaking formulation of
Battjes and Janssen (1978). In the coupling procedure, the wave model (STWAVE) is
run initially using the still-water level and computes the wave radiation stress gradients
which are then used to force the flow (ADCIRC). These wave radiation stress gradients
are under predicted because the setup is not initially considered in the wave model. The
results suggest that more than one iteration between the models is needed to accurately
model the wave setup over a fringing reef. For real life application, this also suggests
that special attention is needed to the coupling between the models when the wave
conditions and water levels are varying rapidly with time and space, such as in the
tropical storm and on reefs to avoid similar problems.
The second laboratory dataset used consisted of the monochromatic waves over a
fringing reef profile typical to that of Hayman Island (Gourlay 1994). The 1D wave
energy balance models were applied to this dataset and produced reasonable agreement
with laboratory measurements. The offset of the computed wave setdown with respect to
measurements indicates that the wave breaking would be better represented by a roller
type formulation for wave breaking (Stive and De Vriend 1994). However, the models
still calculate the wave heights and water elevations within a 50% error.
9.0 Conclusions and Recommendations
Computed wave heights and water levels from two 1D wave energy balance models with
difference wave breaking formulations and the coupled STWAVE and ADCIRC models
were compared with laboratory measurements of combined wind and waves over a
27
fringing reef profile typical of the Island of Guam (Demirbilek et al. 2007a). The three
approaches produce results which are in good agreement with laboratory measurements.
The parametric models, although not intended for quantitative purposes, provide good
qualitative results within a maximum percent error of 50%. These models are intended
for practical applications and not to study the wave mechanics over coral reefs. These
models constitute a useful for tool emergency management and structure design and
other engineering applications.
The most important effect of the wind over coral reefs is the wind setup which allows the
waves to rebuild and increase in height. Due to the shallow depths over fringing coral
reefs, the wind contribution to the setup may be as large as the wave setup especially for
strong winds and wide reef tops.
Acknowledgements. Permission to publish this paper was granted by the Office,
Chief of Engineers, U.S. Army Corps of Engineers. This research was supported by the
Surge and Wave Island Modeling Studies Program.
28
References
Alsina, J.M. and Baldock, T.E. (2007) Improved representation of breaking wave energy
dissipation in parametric wave transformation models, Coastal Engineering, 54,
765-769.
Battjes, J.A., Janssen, J.P.F.M (1978) Energy loss and set-up due to breaking of random
waves, Proceedings Coastal Engineering, ASCE, pp. 569-587.
Brander, R.W., Kench, P.S., Hart, D. (2004) Spatial and temporal variations in wave
characteristics across a reef platform, Warraber Island, Torres Strait, Australia,
Marine Geology, 207, 169-184.
Dally, W.R., Dean, R.G., and Dalrymple, R.A. (1985) Wave height variation across
beaches of arbitrary profile, Journal of Geophysical Research, 90(C6), 11917-11927.
Dally, W.R. (1992) Random breaking waves: field verification of a wave-by-wave
algorithm for engineering applications, Coastal Engineering, 16, 369-397.
Demirbilek, Z., Nwogu, O.G. and Ward, D.L. (2007a) Laboratory study of wind effect on
runup over fringing reefs, ERDC/CHL TR-07-04. Vicksburg, MS: U.S. Army
Engineer Research and Development Center, Coastal and Hydraulics Laboratory.
Demirbilek, Z. and Nwogu, O.G. (2007b) Boussinesq modeling of wave propagation and
runup over fringing coral reefs, ERDC/CHL TR-XX-XX. Vicksburg, MS: U.S. Army
Engineer Research and Development Center, Coastal and Hydraulics Laboratory.
Douglass, S.L. (1990) Influence of Wind on Breaking Waves, Journal of Waterway,
Port, Coastal and Ocean Engineering, ASCE. WW6, 116, 651-663.
Gerritsen, F. (1980) Wave attenuation and wave setup on a coastal reef. Proceedings 17th
International Conference on Coastal Engineering, Sydney, Australia. New York,
ASCE, 444-461.
Gerritsen, F. (1981) Wave attenuation and wave setup on a coastal reef, University of
Hawaii, Look Laboratory, Technical Report No. 48, pp. 416.
Gourlay, M. R. (1996a) Wave set-up on coral reefs. 1. Set-up and wave-generated flow on
an idealized two dimensional horizontal reef. Coastal Engineering 27, 161-193.
Gourlay, M. R. (1996b) Wave set-up on coral reefs. 2. Set-up on reefs with various
profiles, Coastal Engineering, 28, 17-55.
Gourlay, M.R. (2005) Wave-generated flow on coral reefs-an analysis for two-
dimensional horizontal reef-tops with steep faces, Coastal Engineering, 52, 353-387.
Grasmeijer, B.T. and Ruessink, B.G. (2003) Modeling of waves and currents in the
nearshore: parametric vs. probabilistic approach, Coastal Engineering, 49, 185-207.
Janssen, T.T. and Battjes, J.A. (2007) A note on wave energy dissipation over steep
beaches, Coastal Engineering, 54, 711-716.
29
Larson, M., and Kraus, N.C. (1991) Numerical model of longshore current for bar and
trough beaches, Journal of Waterway, Port, Coastal, and Ocean Engineering,
117(4), 326-347.
Longuett-Higgins, M.S. and Stewart, R.W. (1964) Radiation stress in water waves: A
physical discussion with applications, Deep Sea Research, 11.
Luettich, R. A., Westerink, J. J., and Scheffner, N. W. (1992) ADCIRC: An advanced
three-dimensional circulation model for shelves, coasts, and estuaries; Report 1,
theory and methodology of ADCIRC-2DDI and ADCIRC-3DL, ERDC/CHL TR DRP-
92-6, Vicksburg, MS: U.S. Army Engineer Waterways Experiment Station, Vicksburg,
MS.
Lugo-Fernandez, A., Roberts, H.H. and Suhayada, J.N. (1998) Wave transformation
across a Caribbean fringing-barrier coral reef, Continental Shelf Research, 18, 1099-
1124.
Mase, H. (1989) Random wave runup height on gentle slope, Journal of Waterway,
Port, Coastal and Ocean Engineering, 115(5),649-661.
Massel, S.R., and Gourlay, M.R. (2000) On the modeling of wave breaking and set-up on
coral reefs, Coastal Engineering, 39, 1-27.
Miche, M. (1951) Le pouvoir reflechissant des ouvrages maritimes exposes a l’action de la
houle. Annals des Ponts et Chaussessm 12le Anne, 2850319 (translated by Lincoln
and Chevron, University of California, Berkeley, Wave Research Laboratory, Series 3,
Issue 363, June 1954).
Nwogu, O. (2006) Personal communication with Edward Thompson, University of
Michigan, Department of Architecture and Marine engineering, Ann Arbor, MI
48109.
Ruessink, B.G., Walstra, D.J.R., Southgate, H.N. (2003) Calibration and verification of a
parametric wave model on barred beaches. Coastal Engineering, 48 (3), 139–149.
Sánchez, A., Smith, J.M., Demirbilek, Z, Boc, S. (2007) User’s manual to the Coastal
Modeling Package – TWAVE, Unpublished Letter Report, Vicksburg, MS, U.S. Army
Engineer and Research Development Center.
Seelig, W.N. (1983) Laboratory Study of Reef-Lagoon System Hydraulics, Journal of
Waterway, Port, Coastal and Ocean Engineering, ASCE, 109(4), 380-391.
Smith, J.M. (1993) Nearshore wave breaking and decay, ERDC/CHL TR-93-11,
Vicksburg, MS: U.S. Army Engineer Research and Development Center, Coastal and
Hydraulics Laboratory.
Smith, J.M., Sherlock, A.R., Resio, D.T. (2001) STWAVE: Steady-State Spectral Wave
Model: user’s manual for STWAVE, Version 3.0, ERDC/CHL SR-01-01, Vicksburg,
30
MS: U.S. Army Engineer Research and Development Center, Coastal and Hydraulics
Laboratory.
Stive, M.J.F. and De Vriend, H.J. (1994) Shear stresses and mean flow in shoaling and
breaking waves, Proceedings Coastal engineering, ASCE, New York, pp. 594-608.
Thornton, E. B., and Guza, R. T. (1983) Transformation of Wave Height Distribution,
Journal of Geophysical Research, 88(C10), 5925-5938.
Thompson, E.F, and Scheffner N.W. (2002) Typhoon-induced stage-frequency and
overtopping relationships for the commercial port road, Territory of Guam,
ERDC/CHL TR 02-01, Vicksburg, MS: U.S. Army Engineer Research and
Development Center, Coastal and Hydraulics Laboratory.
Thompson, E.F. (2005) A physical model study for investigation of wave processes over
reefs, Coastal Hydraulics Laboratory, Unpublished Letter Report, Vicksburg, MS,
U.S. Army Engineer and Research Development Center.
Van Rijn, L.C. and Wijnberg K.M. (1996) One-dimensional modeling of individual waves
and wave-induced longshore currents in the surf zone, Coastal Engineering, 28, 121-
145.
Van Rijn, L.C., Walstra D.J.R., Grasmeijer, B., Sutherland, J., Pan, S. and Sierra, J.P.
(2003) The predictability of cross-shore bed evolution of sandy beaches at the time
scale of storms and seasons using process-based Profile models, Coastal
Engineering, 47 (3), 295–327.